Recommendations
4 to acquire · 3 to dispose
| Ticker | Name | Close | Action | Action Price Where to act — a buy-limit on a pullback or a buy-stop above resistance, not the current price. Sanity-check vs Close: on a held name, an action price stranded above the close means the queued level was set when the stock was higher — stale, awaiting reprice. | Stop The risk-defined exit, set below the entry for a long (~1.2–1.8×ATR). A held long whose Stop sits at/above the Close is already below its stop — a stale box awaiting reprice. | 1wk Model estimate of the likely price at this horizon — a forward projection, not a guarantee, and separate from the trade's Action Price or Stop. Where the ML estimator is enabled these are per-ticker calibrated forecasts (~80% coverage band) that are refreshed periodically and can drift between refreshes; otherwise they're interpolated from the signal's swing target. Confirm against the settled close. | 4wk Model estimate of the likely price at this horizon — a forward projection, not a guarantee, and separate from the trade's Action Price or Stop. Where the ML estimator is enabled these are per-ticker calibrated forecasts (~80% coverage band) that are refreshed periodically and can drift between refreshes; otherwise they're interpolated from the signal's swing target. Confirm against the settled close. | 12wk Model estimate of the likely price at this horizon — a forward projection, not a guarantee, and separate from the trade's Action Price or Stop. Where the ML estimator is enabled these are per-ticker calibrated forecasts (~80% coverage band) that are refreshed periodically and can drift between refreshes; otherwise they're interpolated from the signal's swing target. Confirm against the settled close. | Rationale |
|---|---|---|---|---|---|---|---|---|---|
| KLAC | KLA Corporation | 172.94 | SELL 30% | 172.94 | 160.55 | 174.15 (+0.7%) | 182.30 (+5.4%) | 179.26 (+3.7%) | Trimming 30% of our KLA position purely as risk discipline. The business is doing fine - record revenue, a raised 2026 industry outlook, and a Street-high target - but our protective stop on this holding has been walked down from to a 41% retreat, while we have never once reduced the position. A risk line that keeps moving away from a holding that never shrinks is a risk failure, so we bank a third of it and keep the rest working. OPERATIONAL: ADR-0020/US-36 Trail Erosion, 40.97% vs peak stop 272.00 (2026-07-01), no size cut since peak -> FIRST fire on this Peak Stop, therefore IMMUNE to US-17/US-18/US-22/US-30/US-34 vetoes; restate this every session or the rule dies. NO signal basis claimed: AVOID label refused (US-29, long-only) and printed Stop_Price 185.82 refused as the bearish-family artifact above a 172.94 close. US-26 declines the PROFIT_ARMED flag as an artifact (Gain_ATR -0.44, -1.60%). Valuation amplify SUPPRESSED by G2 (Est_12W 179.26 / DCF 61.80 = 2.90x, anchor unreliable) so no bump. US-37 STALEMATE: this is the 7TH consecutive unexecuted KLAC card (08-25/26/27/28, 09-01, 09-02, tonight) - requires an explicit TAKEN or SKIPPED Disposition, not an 8th silent re-fire. Residual on stop 160.55 (chandelier 160.06 refused as looser); Peak Stop re-bases on fill. · news |
| NBIS | Nebius Group, N.V. | 210.63 | SELL 30% | 210.63 | 197.16 | 216.66 (+2.9%) | 242.66 (+15.2%) | 302.95 (+43.8%) | Trimming 30% of Nebius on the same risk discipline as KLA - our stop has drifted 19% below its high-water mark while the position size never moved - and with a known supply event five sessions away: Nvidia's warrant over roughly 21 million shares becomes exercisable on 11 September. The forward view is still strongly positive, so we keep working and raise the stop rather than exiting. OPERATIONAL: ADR-0020/US-36 Trail Erosion, peak stop 235.69 (2026-08-17) -> live 191.02 = 18.95%, no size cut since peak -> FIRST fire on this Peak Stop, veto-immune. Secondary: the pre-committed 213.00 line is STILL BROKEN (close 210.63, -1.11%), so the US-22 settled-close self-heal has NOT triggered and the 09-02 auto-withdraw condition did not fire. US-17 respected - the bare NO TRADE cv0 is NOT the basis; L2 does NOT fire (Est_12W +43.83%). Valuation amplify SUPPRESSED by G2 (DCF blank, Rev_Growth 8.55x hyper-growth) so the conservative 30% tier stands. US-31: the 2026-07-05 keep-core overlay is now 60 days stale and carries no news-override force - the core is defended by the live +43.8% forward, not the overlay. Residual, stop RAISED 191.02 -> 197.16 (close - 1.0 x ATR 13.47). AUTO-WITHDRAW: reclaim of 213.00 on the next settled close withdraws this card. · news |
| GOOGL | Alphabet, Inc.- Class A | 342.48 | SELL 30% | 342.48 | 335.34 | 342.95 (+0.1%) | 345.67 (+0.9%) | 385.98 (+12.7%) | Banking 30% of a position that has nearly tripled since we bought it at Alphabet has gone quiet in our model for two straight sessions after reaching 89% of its 12-week objective, and taking a third of a +178% winner off the table while it stalls is ordinary profit discipline. The antitrust picture has actually improved - the court declined to force a Chrome or AdX divestiture and the Street sits at a average target - which is exactly why this is a 30% trim and not an exit: stay long on a raised stop. OPERATIONAL: ADR-0012 armed + fired. ARM VERIFIED under US-26 via the hard Gain_ATR limb - Gain_ATR 30.75, Unreal_PnL_Pct +178.44%, both > 0 (contrast KLAC, artifact-declined tonight). FIRE = tick (d), 2 consecutive NO TRADE prints (09-02, 09-03). Ticks (a) no second Exit_Warning code, (b) cv is the NoTrade structural artifact and is DECLINED as a measured deterioration (US-29), (c) Target_12Week 385.98 > 1.02 x 342.48 = 349.33 so the forward is ALIVE - no fire. US-33 does not bite: the arm comes from the Gain_ATR limb, not the T12 limb, and the fired tick is (d) not (c). US-19: soft-tick-only fire -> execute at the 30% tier, do NOT escalate; the 09-01 fire was withdrawn 09-02 when tick (d) genuinely broke and has now been re-verified on fresh evidence, which is a new fire, not an escalation. US-34 noted and unaffected - no STRONG SELL - REDUCE print is being relied on. Valuation: MoS -0.631 > -3, no tier bump. Residual, trail RAISED 333.68 -> 335.34 (close - 1.0 x ATR 7.14). · news |
| DELL | Dell Technologies Inc | 516.39 | BUY | 514.73 | 395.13 | 520.88 (+1.2%) | 564.97 (+9.8%) | 649.92 (+26.3%) | The strongest reading in the entire book: a maximum-conviction 52-week breakout, ranked first of 51 names, on a company that just booked a record 9bn of AI server orders, exited the quarter with a $95bn backlog and lifted FY27 guidance to $192bn of revenue. It is one of only three names here that still passes our quality screen, with an F-Score of 8, and it trades near our own fair-value estimate rather than far above it. We buy at the breakout trigger. OPERATIONAL: cv100, Buy_Rank 1, Entry_Distance_ATR -0.04 = AT ENTRY, so the no-chase bar is not tripped. REPRICED from the unfilled 09-02 card at 492.20 per the action-price provenance rule - DELL ran +4.91% unfilled, US-9 forbids carrying the stale level forward, and the model has re-armed its own trigger at 514.73. L6/US-9 CONFLICT STATED AND OVERRIDDEN: this is session 2 after a +15.81% earnings gap; the override rests on the at-entry distance, cv100, Buy_Rank 1 and Quality_Pass True, and is expressed through the 0.75x size modifier (MoS -2.768) rather than through a fuller position. Quality gate PASSES (F=8, Bargain True). Scorecard BreakoutUp_52W|buy +0 (n=10, actionable) - notably NOT one of the three momentum-chase trap families under US-16. Sizing: Portion 85.49 -> 0.8549 x 200000 x 20% =; x0.75 =; / 514.73 = = (4.1% of the marked book). Est_12W 649.92 vs DCF 485.61 = 1.34x - noted, but FAIR_VALUE not GROWTH_PRICED_IN so no formal annotation. PRE-COMMITTED WITHDRAW: family leaves BreakoutUp_52W; conviction < 65; Quality_Pass flips False; a settled close below 395.13; or a 09-04 open gapping >1.5% above 522.45 (reprice at the next EOD, do not chase). · news |
| MSFT | Microsoft Corp | 510.12 | BUY — QUEUED @498.20 | 498.20 | 484.80 | 512.31 (+2.8%) | 542.44 (+8.9%) | 552.15 (+10.8%) | Microsoft on the one entry pattern that has actually worked for this book - a pullback to its volume-weighted average price - upgraded to a top-conviction reading after a 9% August on strong cloud and AI guidance, with Bank of America at a target. It clears our quality screen, which most of today's candidates do not. The stock has already run 1.0 ATR past the model's entry, so we do not chase it: we leave a resting order at and only buy on a pullback. OPERATIONAL: L1-favoured family. cv ->86 (fresh upgrade, not a same-day flip). Entry_Distance_ATR -1.03 = EXTENDED past entry -> QUEUED GTC limit per the no-chase rule, NOT market. Quality gate PASSES (Q_Pass True, F=6). Scorecard PullbackBuy_VWAP|buy +1 (n=24, actionable) - note US-38/US-24: the cell has eroded from +5 to +1 and its edge is 1wk-only, so treat Target_Price 525.02 as the risk objective, not an expectation. Sizing: Portion 85.49 ->; x0.75 (MoS -1.747) =; / 498.20 = = Est_12W 552.15 vs DCF 228.73 = 2.41x - WARNING: exceeds DCF anchor, the projection assumes growth beyond what is already priced in. expires_on 2026-09-09 (anchor + 3 sessions; 09-07 is Labor Day). PRE-COMMITTED WITHDRAW (US-8): family leaves PullbackBuy_VWAP, conviction < 65, Quality_Pass flips False, or expiry - auto-withdraw, do not reprice upward into strength. · news |
| PLTR | Palantir Technologies Inc. | 182.53 | BUY — QUEUED @177.36 | 177.36 | 165.94 | 184.34 (+3.9%) | 188.79 (+6.4%) | 177.28 (-0.0%) | Palantir on the same pullback pattern as Microsoft, after a session that jumped 7.7% on a new US Army TITAN contract and an expanded PwC partnership, with revenue up 93% and US commercial up 149%. The catch is price: it trades at a trailing P/E near 255 and is the most expensive name in this book against our own valuation work, so we take it at half our normal size and only on a pullback to - and we note that our 12-week model sees no upside from that level at all. OPERATIONAL: L1-favoured family, cv ->86 strengthening. Entry_Distance_ATR -0.52 = extended -> QUEUED GTC limit, not market. Quality gate PASSES (Q_Pass True, F=8, Quality_Rank 7). val_adj: MoS -13.405 x 5 = -67 CLAMPED to -10 - the clamp is doing heavy lifting here and the size modifier is the real expression of the valuation risk: MoS <= -3 -> x0.50. Sizing: Portion 85.49 ->; x0.50 =; / 177.36 = = FLAG: Est_12W 177.28 is -0.05% vs the action price - a DEAD 12-week forward on a STRONG BUY row (the same pattern that sank the 08-28 PLTR queue at 166.99). The 1wk +3.94% and 4wk +6.44% carry this trade; the 12wk does not. Est_12W 177.28 vs DCF 46.62 = 3.80x - WARNING: exceeds DCF anchor, projection assumes growth beyond what is priced in. expires_on 2026-09-09. PRE-COMMITTED WITHDRAW: family flip off PullbackBuy_VWAP, conviction < 65, Quality_Pass False, or expiry. · news |
| AAPL | Apple | 328.21 | BUY-STOP — QUEUED @329.13 | 329.13 | 307.36 | 326.47 (-0.8%) | 319.52 (-2.9%) | 330.38 (+0.4%) | A half-size starter in Apple that only triggers if the stock breaks higher on its own - a resting buy-stop at ahead of the 09-09 product event. Apple has the best fundamental quality score in the entire book (F-Score 9) and has now signalled accumulation for three straight sessions with conviction rising to its highest yet. We keep it small and stop-triggered because our own model sees the shares roughly flat over the next four weeks, so we want the market to confirm the move before we pay for it. OPERATIONAL: BUY - ANTICIPATE in 3 of the last 3 snapshots (09-01 cv83 -> 09-02 cv77 -> 09-03 cv88), well clear of the 55 slot floor. REPRICED 328.04 -> 329.13 per provenance; the prior GTC is NOT carried forward. Buy-stop sits 0.28% ABOVE the 328.21 close, so it fills only on confirmed strength - the S3 anomaly did not recur tonight. The 09-02 PRE-COMMITTED WITHDRAW TRIGGER was CHECKED AND DID NOT FIRE: it required Est_4W below the action price AND conviction < 70 - Est_4W 319.52 < 329.13 is true, but cv 88 >= 70 is not, so the order stands (US-19 bars killing it on an ad-hoc basis). Quality gate PASSES. Scorecard AnticipationUp_VCP|buy n=4, REPORT-ONLY, not applied. Sizing: Portion 60.76 ->; x0.75 (MoS -2.975) =; x0.5 anticipate slot =; / 329.13 = = (1.4% of book). Est_12W 330.38 vs DCF 237.39 = 1.39x - WARNING: exceeds DCF anchor. ADVISORY, CITED NOT APPLIED (Step 4.4 / US-28): counterfactuals.anticipate is now n=9, median -3.46% at 1wk / -9.86% at 4wk [saved] - this book has been consistently better off NOT taking anticipation entries. Never blended into conviction; it goes to the learnings review. Retained because US-20/US-21 hold that the buy-STOP is this book's only working entry lever during the quality-gate outage. expires_on 2026-09-09. |
| LLY | Eli Lilly & Co | 1159.60 | HOLD | 1159.60 | 1168.46⚠ | 1161.00 (+0.1%) | 1185.12 (+2.2%) | 1152.42 (-0.6%) | Holding all, but this is now the most deteriorated position in the book and it is one step from a trim. Lilly has closed below our protective stop for a third straight session, has given our model no signal at all for three sessions, and as of tonight the 12-week outlook has turned mildly negative for the first time. What holds us is the line we committed to in advance: the decisive level is and the stock is still above it, and the shares have been flat, not falling. OPERATIONAL: L2's SECOND LEG NEWLY SATISFIED - Est_12W 1152.42 is -0.62% vs the 1159.60 close (was +1.94% on 09-02). The stop-break leg is NOT decisive: 0.76% below the 1168.46 stop, and the PRE-COMMITTED decisive line 1142.00 is intact. US-27 governs - honour the pre-committed line, do not ad-hoc override it. ADR-0012: PROFIT_ARMED via the T12_Progress limb ONLY (100.62). Ticks (c) and (d) technically fire (Target_12Week 1152.42 < 1.02 x 1159.60 = 1182.79; three consecutive NO TRADE prints 09-01/02/03) BUT: US-33 DISQUALIFIES (c) as the same metric the arm reads, and US-26 CANNOT BE VERIFIED - Avg_Cost is blank in input/usportfolio.json so Gain_ATR and Unreal_PnL_Pct come back null and the arm cannot be confirmed real. Declined to originate rather than guessing (systemic finding S7, third consecutive session). PRE-COMMITTED: a settled close below 1142.00 completes L2 and fires a 30% trim at the next EOD without further deliberation. Stop 1168.46 unchanged. |
| NVDA | NVIDIA Corporation | 228.45 | HOLD | 228.45 | 220.48 | 230.30 (+0.8%) | 231.81 (+1.5%) | 236.14 (+3.4%) | Holding all and raising the trailing stop to. Nvidia printed the strongest signal of any holding tonight - conviction jumped from as the shares rose 1.8% - and the sensible response to a position that is both winning and strengthening is to protect more of the gain, not to cut it. OPERATIONAL: ADR-0012 ARMED-BUT-NOT-FIRED -> mandatory trail-raise, not a trim. Arm reads PROFIT_ARMED via T12_Progress 96.74. Tick (a) RSI_DIV is present, which would ordinarily fire a 30% trim - but US-26 blocks origination: Avg_Cost is blank so Gain_ATR and Unreal_PnL_Pct are null and the arm cannot be verified real. Declined by rule, consistent with 09-01/09-02 (systemic finding S7). Ticks (b) conviction ROSE (the opposite of deterioration), (c) Target_12Week 236.14 > 1.02 x 228.45 = 233.02 so the forward is alive, (d) HOLD x4, no NO TRADE. Trail RAISED 217.07 -> 220.48 (close - 1.0 x ATR 7.97); model Stop_Price 215.33 and the 2.0x ATR chandelier 212.51 are BOTH refused as looser (US-29). PRE-COMMITTED: once Avg_Cost is backfilled, re-adjudicate the RSI_DIV tick at the next EOD - this is the third session a legitimate profit-take has been blocked by a one-line data gap. |
| MU | Micron Technology, Inc. | 958.16 | HOLD | 958.16 | 929.07 | 971.57 (+1.4%) | 1131.00 (+18.0%) | 1345.91 (+40.5%) | Holding all. Micron continues to strengthen quietly - conviction rose from - it sits 11.5% above our cost, 3.0% clear of its trailing stop, and carries the most positive long-term outlook of anything we own. Nothing to do. OPERATIONAL: NOT armed - Gain_ATR 2.37 < the 3.0 threshold and T12_Progress 71.19 < 90, no Exit_Warning. Trail 929.07 held: model Stop_Price 918.88 and close - 1.0 x ATR (916.48) are both refused as LOOSER (US-29 variant - never consume a stop that widens risk). Est_12W +40.5% is the book's best forward. |
| OKTA | Okta Inc. | 170.42 | HOLD | 170.42 | 163.59 | 172.69 (+1.3%) | 180.22 (+5.8%) | 194.45 (+14.1%) | Holding all and tightening the stop to. Okta was yesterday's only loser and today's second-best gainer, up 4.5%, recovering to a live buy-family signal and 22% above our cost. We use the strength to raise protection. OPERATIONAL: recovered NO TRADE -> HOLD cv63. NOT armed (Gain_ATR 2.90 < 3.0, T12_Progress 87.64 < 90, no Exit_Warning) - note how close both limbs are; a further move arms it next session. Trail RAISED 160.20 -> 163.59, adopting the model Stop_Price because it is TIGHTER than the standing trail (the one case where the model stop is safe to consume on a held long). 4.0% clear of the new line. |
| NET | Cloudflare, Inc. | 284.51 | HOLD | 284.51 | 283.72 | 286.04 (+0.5%) | 284.67 (+0.1%) | 296.98 (+4.4%) | Holding all - the risk event we flagged last night has resolved. Cloudflare rose 4.3% and closed back above our protective stop, so the breach healed without us having to act. OPERATIONAL: close 284.51 is +0.28% vs the stop (was -3.87% on 09-02); the pre-committed decisive line 270.58 was never approached. L2 does not fire (Est_12W +4.38%). Stop 283.72 unchanged - the margin is razor-thin at 0.28%, so a single down session re-opens the case. DATA DEFECT PERSISTS (S9, third session): input/usportfolio.json still carries TWO NET rows ('Clouadflare Inc' / 'Cloudflare, Inc.'); the scan therefore reads Held=N and the monitor again reported 'unledgered drift' instead of tracking the position. Dedupe the holdings file and run swingtrade2.py trade reconcile. |
| SPCX | Space Exploration Tech Corp | — | HOLD | — | 129.00 | n/a | n/a | n/a | Holding all on the manual stop. No thesis problem, but a dated supply event lands in six days: another 7% tranche of insider shares unlocks around 10 September, part of a staggered schedule that runs in 7% blocks until Musk's 6.4 billion shares in June 2027. The first two unlocks in August were absorbed without damage, which is the encouraging part. Watch, do not act. OPERATIONAL: MUTED ON THE VERDICT - absent from the scan CSV, so no Action, conviction, entry or targets are synthesized. NEWS-ASSESSED as required by the unpriceable-holding rule: classification 'watch'. S8 REMAINS OPEN (third session): the permanent-mute rationale ('private, no public quote') has EXPIRED - SPCX trades on Nasdaq - yet it is still outside the scan universe, so a 300-share position runs with no signal, no automated stop check and no fusion treatment. Add it to the universe. AUTO-UNMUTE: resumes full signal + fusion treatment the moment it appears in the scan CSV with a non-null Close. · news |
Outcome & track record
Accuracy at the time of this record.
180 Graded decisions Past recommendations old enough to score against what actually happened (target hit / stopped out / closed). This is the evidence base — the bigger the number, the more history behind the calls. Click to inspect every one. See the decisions →
-0.191 1-week calibration How much of the eventual 4-week move typically shows up in the first week (0 = almost none, 1 = most of it). Use it to read the 1-week estimates: a low number means prices drift toward the target slowly, so don't over-read a soft week-1.
52 Published records How many dated end-of-day verdicts exist for this book — the length of the public track record you can browse in the Archive below.